I am a postdoctoral researcher in the Mathematics for Materials Modelling group of Prof. Michael F. Herbst at EPFL, where I work on machine learning for electronic structure simulations, in particular neural operators for accelerating and improving the accuracy of density functional theory. Before that, I completed my PhD in machine learning at the University of Tübingen and the IMPRS-IS with Prof. Philipp Hennig on probabilistic numerical methods for ordinary differential equations; see my PhD thesis "A Flexible and Efficient Framework for Probabilistic Numerical Simulation and Inference". More broadly, I'm interested in how machine learning can make scientific simulations faster, more accurate, and more reliable, which includes topics such as probabilistic numerics, surrogate modeling, state-space models, differential equations, Bayesian optimization, and Gaussian processes.
I like to make my research widely accessible in the form of open-source software: ProbNumDiffEq.jl provides efficient probabilistic numerical differential equation solvers in Julia that are compatible with the broader DifferentialEquations.jl ecosystem — it contains nearly all the methods that I developed during my PhD. I also contributed to the probnum probabilistic numerics toolkit and maintain a range of smaller Julia packages, such as TuePlots.jl for making publication-ready plots with Makie. Check out the code section!